Enhancing Cyclist Safety Through Driver Gaze Analysis at Intersections With Cycle Lanes

Abstract

In urban areas, roads with dedicated cycle lanes play a vital role in cyclist safety. However, accidents can still occur when vehicles cross the cycle lane at intersections. Accidents mostly occur due to failure of the driver to see a cyclist on the cycle lane, particularly when the cyclist is going straight through the intersection, and the vehicle is turning. For safe driving, it is critical that the drivers visually scan the area in the vicinity of the junction and the car, particularly using the wing-mirror, prior to making turns. This paper describes results from a set of test drives using in-vehicle non-invasive eye-tracking and in-vehicle CAN bus sensors to determine driver behaviour. In total, 20 drivers were monitored through 5 different intersections with cycle lanes. The study found that approximately 83% of drivers did not check their wing mirror prior to, or during their turning manoeuvre, potentially putting pedestrian, cyclists, scooter and hoverboard users in danger. An algorithm was developed to analyse driver gaze during the turning manoeuvre to identify cases where they failed to look at the wing mirror. The gaze pattern and gaze concentration on the mirror helps to identify safe and unsafe driving behaviour. This information can then be used to improve Advanced Driver-Assistance Systems (ADAS) to create a safer environment for all road users.